The Preventive Use of Force: A
Cosmopolitan Institutional Proposal
Allen Buchanan and Robert O. Keohane *
ince September 11 fears of terrorism and
S weapons of mass destruction have fueled
a vigorous worldwide debate about the
preventive use of force. Preventive use of force
may be defined as the initiation of military
action in anticipation of harmful actions that
are neither presently occurring nor imminent.1
This essay explores the permissibility of
preventive war from a cosmopolitan norma-
tive perspective, one that recognizes the
basic human rights of all persons, not just
citizens of a particular country or countries.
We argue that within an appropriate rule-
governed, institutional framework that is
designed to help protect vulnerable coun-
tries against unjustified interventions with-
out creating unacceptable risks of the costs
of inaction, decisions to employ preventive
force can be justified.
In our proposal, states proposing preven-
tive war would have to enter into a contract
with a diverse body of states as a condition for
authorization of their actions. Both propo-
nents of action and those opposing it would
be held accountable, after the fact, for the
accuracy of their prior statements and the
proportionality of their actions. This institu-
tional arrangement is designed to improve the
quality of decisions on the preventive use of
force. It does so by insisting that these deci-
sions take place under rules that create incen-
tives for honest revelation of information and
responsible conduct, and that they be made by
agents that are comparatively morally reliable.
The key to ensuring the fairness of rules
governing the preventive use of force is
accountability. Our proposed scheme pro-
motes accountability through a combination
of ex ante and ex post mechanisms. Prior to
taking preventive action, states will be
required to enter into a contingent contract
that imposes two requirements. First, they
must make an evidence-based case for pre-
ventive force to the UN Security Council. Sec-
ond, they must agree in advance to submit
themselves to an evaluation by an impartial
body after the preventive action occurred.
*
The authors are grateful to J. L. Holzgrefe for stimulat-
ing conversations that helped to generate this paper. We
have benefited from comments on presentations of this
paper from participants at colloquia held at the Woodrow
Wilson School, Princeton University, February 19, 2003;
the Faculty Colloquium, Sanford Institute of Public Pol-
icy, Duke University, February 24, 2003; Duke Law School,
March 19, 2003; the University of Sydney Law School,
August 22, 2003; Columbia University Law School, Octo-
ber 28, 2003; and the University of California Law School,
November 17, 2003. We thank Michael Byers, Neta Craw-
ford, Ruth Grant, Henry R. Nau, Joseph S. Nye Jr., and two
anonymous reviewers for valuable written comments on
an earlier draft. We are particularly grateful to the editors,
Christian Barry and Joel Rosenthal, for their extensive and
perceptive suggestions. We are grateful for research sup-
port from the Triangle Institute for International Secu-
rity–Duke University project “Wielding American
Power,” funded by the Carnegie Corporation of New York
and its International Peace and Security Program.
1
The notion of an “imminent”attack is somewhat vague but
is usually understood to include not only situations in which
missiles or warplanes have been launched but have not yet
struck their targets, but also situations in which forces have
been mobilized with apparently aggressive intent.
1
This body will be charged with determining
whether the empirical claims that were
employed to justify the preventive action were
true. If the ex post evaluation undermines the
intervening states’ justification for acting, the
contract would provide for sanctions against
them. If the ex post evaluation vindicates the
recourse to preventive force, then the contract
would impose sanctions on those members of
the Council who opposed the proposed
action. If preventive action were blocked by a
majority vote of the Security Council or a veto
by one of the permanent members, those
seeking to engage in preventive action could
then make their case in a different body—a
coalition of democratic states—with its own
institutionalized mechanisms for accounta-
bility.Although our goal is to develop an insti-
tutional framework for decisions concerning
the use of preventive force, we believe that our
general approach can also be used to develop
a framework for making decisions concerning
humanitarian military intervention to stop
presently occurring massive violations of
basic human rights. We focus more narrowly
on preventive force for two reasons: the pre-
ventive use of force involves special risks and
is thus more difficult to justify; and the Bush
administration’s recent claims that the right of
self-defense includes the permission to engage
in the preventive use of force has made this of
urgent practical importance.
FOUR VIEWS ON THE PREVENTIVE
USE OF FORCE
Four distinct positions in the current debate
on preventive force have emerged: the Just
War Blanket Prohibition; the Legal Status
Quo; the National Interest; and the
Expanded Right of Self-Defense. Clarifying
these views and identifying their shortcom-
ings helps to illuminate the distinctive fea-
tures of our proposal.
2 Allen Buchanan and Robert O. Keohane
The Just War Blanket Prohibition. The dom-
inant view in the just war tradition has been
that preventive force is strictly forbidden.
Force may sometimes be justified in cases in
which an attack has not already occurred but
is imminent—when, for example, an enemy is
mobilizing his forces with clear aggressive
intent or when missiles or warplanes have
already been launched but yet not struck their
targets––but there is generally thought to be a
blanket prohibition on preventive action.2
The Legal Status Quo. States’ preventive
use of force is generally regarded as prohib-
ited in contemporary international law
unless they have received collective authori-
zation by the UN Security Council. Article
2(4) of the UN Charter requires “all states”
to refrain “from the threat or use of force
against the territorial integrity or political
independence of any state,” unless author-
ized by the Security Council (Articles 39, 42,
48) or in self-defense against an armed
attack (Article 51). According to the Legal
Status Quo view, this highly constrained
stance on preventive force ought to be main-
tained. Preventive force should be used only
with Security Council authorization.3
2
Michael Walzer, Just and Unjust Wars (New York: Basic
Books, 1977).
3
In the absence of authoritative judicial determination,
international law is subject to a variety of interpretations.
Some might hold that under current international law
preventive action can be legitimate because the Genocide
Convention obligates states/parties to take action to pre-
vent genocide. However, this claim is contestable
because, according to Article 103 of the UN Charter, the
prohibition against preventive use of force trumps all
other treaties, including the Genocide Convention. It
might also be held, on the contrary, that preventive force
is not permissible under international law even with
Security Council authorization, except in order to
respond to a “threat to international peace and security.”
For our purposes, it is not necessary to take a stand on
these legal questions. We merely follow most authorities
in treating the legal status quo as prohibiting preventive
use of force without Security Council authorization, but
permitting it if such authorization has been provided.
 The National Interest. Realists hold that
states may do whatever their leaders deem
necessary to serve the best interests of the
state. According to this view, leaders of states
may disregard universal moral principles
when they conflict with the national inter-
est. More specifically, they may employ
force, including preventively, if they deem it
necessary for the pursuit of state interests.4
The Expanded Right of Self-Defense. The
Bush administration’s “National Security
Strategy” articulates a fourth position. It
expands the definition of self-defense to
include preventive action: “While the United
States will constantly strive to enlist the sup-
port of the international community, we will
not hesitate to act alone, if necessary, to exercise
our right of self-defense by acting preemptively
against such terrorists, to prevent them from
doing harm against our people and our coun-
try [emphasis added].”5 The Bush adminis-
tration’s “National Security Strategy” asserts
that states possess a right of self-defense that
entitles them to take preventive action.
None of these four views provides an ade-
quate model for governing the preventive
use of force. Adherence to the Just War Blan-
ket Prohibition is too risky, given the wide-
spread capacity and occasional willingness
of states and nonstate actors to deploy
weapons of mass destruction covertly and
suddenly against civilian populations. In
addition, this view requires states to refrain
from acting even when they could prevent
massive human rights violations at little
cost.6 The National Interest view is also
unacceptable. Indeed, it repudiates all
progress that has been made in constraining
the international use of force. The problem
is not simply that the national interest is
such a malleable concept that its invocation
is likely to provide a rationale for aggression
and a recipe for destructive international
instability, but that by reducing the grounds
the preventive use of force
for the preventive use of force to self-
defense, this view conceives of the potential
aims of preventive action too narrowly. It
fails even to consider the possibility that
there are circumstances in which the pre-
ventive use of force would be justified to
protect the rights of persons other than
one’s fellow citizens.
The Expanded Right of Self-Defense view
must also be rejected. Allowing states to use
force on the basis of their own estimate that
they may be attacked in the future, without
provision for checks on the reliability and
sincerity of that judgment, would make the
use of force too subject to abuse and error.
Like the National Interest view, it also focuses
only on the welfare of those within one’s own
state, thereby providing no basis for the pre-
ventive use of force on behalf of others.
The Legal Status Quo view is quite attrac-
tive when compared to the three views
described above, since it allows preventive
action, including its use to protect persons
beyond one’s borders, while also providing
checks on the judgment of any particular
state or group of states seeking to use it. But
when compared to a fifth alternative—the
Cosmopolitan Institutional view—the
Legal Status Quo view appears far less com-
4
For the classic statement, see Hans J. Morgenthau, Pol-
itics Among Nations: The Struggle for Power and Peace,
6th ed., revised by Kenneth W. Thompson (New York:
Knopf, 1985), p. 12. Advocates of the National Interest
view need not be moral skeptics; instead they may hold
that state leaders have one supreme moral obligation—
to serve the national interest—and that this obligation
overrides all other principles of morality. Those who
espouse the National Interest view may disagree among
themselves, of course, as to whether a policy of recourse
to preventive force, or any particular decision to use
preventive force, is in fact likely to serve the national
interest or be detrimental to it.
5
“National Security Strategy of the United States
of America September 2002,” p. 6; available at www.
whitehouse.gov/nsc/nss.pdf.
6
We explore the flaws of the Just War Blanket Prohibi-
tion in more detail below.
3
pelling. The normative foundation of the
Cosmopolitan Institutional view is cosmo-
politan in the sense that it takes the human
rights of all persons seriously. By basic
human rights we mean the most widely
acknowledged rights that are already recog-
nized in the major human rights conven-
tions. These include the right to physical
security of the person, including the right
against torture, and rights against at least
the more damaging forms of discrimina-
tion on grounds of religion, gender, race, or
ethnicity, as well as rights against slavery,
servitude or forced labor, and the right to
the means of subsistence.7 A commitment
to protecting and preventing massive viola-
tions of human rights provides a prima
facie justification for preventive force as a
last resort.
These commitments of the cosmopolitan
perspective are far from novel. Indeed, they
are central to the just war tradition and the
current international legal order’s allowing
human rights to limit state sovereignty. The
distinctiveness of the Cosmopolitan Institu-
tional view lies in how it incorporates these
normative commitments with an effective
accountability regime for responsible deci-
sion-making concerning the preventive use
of force.
A cosmopolitan accountability regime
involves both substantive standards and
processes. The substantive standards
include traditional just war principles,
including competent authority, propor-
tionality, noncombatant immunity, realis-
tic likelihood of success, and avoidance of
excessive force against enemy combatants.
Unfortunately, the chief deficiency of just
war theory lies not in its principles but in
the failure of the just war theorists to take
seriously the need for institutionalizing
them in a way that provides incentives for
agents to apply them impartially and on
4 Allen Buchanan and Robert O. Keohane
the basis of good information. Current
international law does institutionalize the
important just war principle of competent
authority: the requirement that the use of
force must be authorized by the Security
Council, except in cases of self-defense
(when there is no time to consult the
Council).8 And in this sense, the Legal Sta-
tus Quo view can be seen as an important
advance from an anarchic situation in
which states unilaterally decide whether to
use force. But this way of institutionaliz-
ing the use of force is seriously inade-
quate, at least when it comes to the
preventive use of force.
THE COSMOPOLITAN JUSTIFICATION
FOR PREVENTIVE WAR
We begin with the assumption that it can be
morally permissible to use force to stop
presently occurring massive violations of
basic human rights. We then argue from this
assumption that there is at least a prima
facie case for the moral permissibility of
7
The position we endorse is sometimes called Moder-
ate Cosmopolitanism, which allows one to give a lim-
ited priority to the interests of one’s own nation and
does not require strict impartiality. It is a liberal form
of cosmopolitanism, since it emphasizes the basic
human rights of all persons. One could imagine other
kinds of cosmopolitanism, such as Marxism, that for-
bid discrimination on grounds of nationality, but do
not focus on human rights. For a philosophical justifi-
cation for the assumption that there are human rights,
see Allen Buchanan, “Human Rights,” in Justice, Legiti-
macy, and Self-Determination: Moral Foundations for
International Law (forthcoming, Oxford: Oxford Uni-
versity Press, 2003).
8
The humanitarian law of war can be seen as an imper-
fect institutionalization of some of the most important
jus in bello principles of just war theory. However, given
the distinctive risks of the preventive use of force, spe-
cial institutional arrangements are required for respon-
sible decisions to use force preventively, over and above
better compliance with rules of humane warfare that
are designed to apply to all uses of force.
using force to prevent massive violations of
basic human rights.9 The core justification
for using force to stop rights violations as
they are occurring—the need to protect
basic human rights—can also justify the use
of force to prevent rights violations.10
Reflection on two scenarios will help to
flesh out the prima facie case for this propo-
sition. In the first case, a group is already in
the process of releasing a weaponized,
extremely virulent, lethal virus into the
heart of a major city. Surely, it would be jus-
tifiable to intervene forcibly to stop them
from releasing more of the deadly concoc-
tion. In the second scenario, the intelligence
agencies of a state have consistent informa-
tion from many reliable sources that a group
that has deliberately killed civilians in the
past has in its possession a weaponized,
extremely virulent, lethal virus and plans to
use it against a civilian population. Suppose
further that the current location of the virus
is known but the city at which it is targeted
is not, and that once the virus leaves its pres-
ent location it will be very difficult to track.
Let us also stipulate that the current location
of the virus is in the remote stronghold of
the group and that a preventive strike can
destroy the virus without killing any persons
who do not belong to the group. Under these
circumstances, the need to protect basic
human rights supports preventive action,
regardless of whether the rights endangered
are those of our fellow citizens or foreigners.
Reflection on such examples supports the
conclusion that preventive action can be
ethically permissible.
It is crucial to emphasize that this prima
facie justification for the preventive use of
force does not apply to all cases where harm
may be prevented but only to situations in
which there is a significant risk of sudden
and very serious harms on a massive scale.
Such a risk is inherent in weapons of mass
the preventive use of force
destruction but not exclusive to them.
Genocides may also erupt suddenly.
In situations characterized by incremen-
tally increasing violence, such as in the for-
mer Yugoslavia, however, the case for
preventive war is much less compelling.
Action to respond to aggression or acts of
ethnic violence can feasibly be taken in such
contexts with far less uncertainty regarding
the need to act after the first human rights
violations have already occurred. Thus the
force of the prima facie argument is limited
and focused. It does not purport to show that
force may be used whenever it is likely to stop
massive violations of basic human rights.
What the Cosmopolitan Institutional
view and the Legal Status Quo view have in
common is the assertion that there are
some cases in which preventive action
ought to be permitted. The prima facie
argument for prevention sketched above
provides a needed moral foundation for
both of these positions.
A Critique of the Just War Blanket
Prohibition Argument
Those who endorse the blanket prohibition
characteristic of the just war tradition could
point out that the prima facie argument
9
The prima facie argument seems to imply that preven-
tive action to protect basic human rights is not only per-
missible but also obligatory. However, since we seek to
advance a feasible proposal, our discussion of institu-
tional arrangements does not assume such an obliga-
tion. The successful implementation of the
accountability regime we propose might be an impor-
tant step toward the more ambitious goal of institution-
alizing an obligation to prevent massive violations of
basic human rights.
10
By beginning with the assumption that the use of
force is sometimes permissible to stop massive viola-
tions of basic human rights already under way, we are
of course implicitly rejecting the absolute pacifist view
that the use of military force is never justified because
it inevitably involves harm to innocent persons. We
share this assumption with the four perspectives that
we criticize.
5
ignores a crucial point about preventive
force: it is directed toward someone who has
not committed a wrong. Thus preventive
action violates the rights of its target. Unless
this moral objection can be met, both the
Cosmopolitan Institutional view and the
Legal Status Quo view must be rejected.
This objection is unconvincing, since it
rests on a false assumption about the right
not to be attacked. There is, trivially, a right
not to be attacked unjustly. But to assume
that this includes the right not to be attacked
unless one has already committed, or begun
to commit, a wrongful harm is to beg the
question at issue—namely, whether preven-
tive force can ever be morally justified.
Reflection on the right of self-defense
suggests that preventive action need not vio-
late the rights of the target. At common law,
an individual may use deadly force in self-
defense if a reasonable person in his circum-
stances would judge that he is in danger of
death or serious bodily injury. Of course,
other conditions must be fulfilled as well,
including the avoidance of excessive force.
The point, however, is that there are circum-
stances in which it is justifiable to use lethal
force even if the target against which it is
applied has not yet caused harm. If this is the
case, then it may also be permissible to use
lethal force in cases where a reasonable actor
would judge that the behavior of another
person or persons pose a serious threat of
death or serious bodily harm even if an
attack has not yet begun.
In the case of individuals, there is much to
be said for restricting the legal right of self-
defense so as to require the reasonable belief,
by the agent exercising the right of self-
defense, that an attack is already under way.
This restriction is reasonable in contexts in
which there are a well-functioning police
force and legal system that serve to reduce
the risk that individuals will be attacked by
6 Allen Buchanan and Robert O. Keohane
their fellow citizens. In the international
arena there is nothing comparable to an
effective police force, and the stakes are often
much higher, since millions of people may
be killed by a sudden deployment of
weapons of mass destruction. Under these
circumstances the restrictive interpretation
of the right to self-defense seems inade-
quate. It seems much more plausible to
assert that the use of force to prevent great
harm can be justified, even if that harm is
neither presently occurring nor imminent.
In the second scenario described above,
the terrorist group has not yet launched an
attack, and the harm it seeks to inflict is not
literally imminent. Yet it is incorrect to say
that the group has done nothing. It has
wrongfully imposed an especially high risk of
serious harm on others, through its past
actions, its current planning to carry out a
lethal attack, and its expression of a willing-
ness to kill noncombatants. The group’s
actions, plans, and avowed intentions have
put innocent people at risk for great harms.
Reflection on the law of conspiracy sug-
gests that using force against someone who
has not yet committed a wrongful harm need
not violate his rights. The elements of con-
spiracy include a “specific intention” to do
wrongful harm and an “agreed plan of
action” to produce the harm. The “specific
intention” requirement rules out mere unfo-
cused malevolence as a trigger for criminal
liability, while the requirement of an “agreed
plan of action” satisfies the condition that a
crime must include an act, not merely a
guilty mind. There is nothing morally repug-
nant about the idea of using force against
conspirators simply because they have not
yet performed the act they have planned. We
can dismiss the simplistic allegation that it
cannot be right to use force against him
because “he hasn’t done anything” once we
acknowledge that the conspirator has indeed
done something by agreeing to a plan to pro-
duce wrongful harm.
One plausible explanation of why it can
be justifiable to use force against someone
who has a “specific intention” and an
“agreed plan of action” to do wrongful harm
is that he has wrongfully imposed a risk of
serious harm on others. It is a wrongful
imposition of risk if those put at risk have
neither voluntarily accepted the risk nor
deserve to be subjected to it. The crucial
point is that when someone has wrongfully
imposed a risk, it can become morally per-
missible to do things to alleviate the risk that
would otherwise be impermissible. This is
not to say that one may do whatever is nec-
essary to alleviate the risk. As with the use of
force generally, the principles of propor-
tionality, avoidance of excessive force, and
demonstration of proper regard for the
rights of innocent persons must be
observed. Whether the use of preventive
force is justified in such cases will depend, as
with self-defense against an imminent
attack, upon whether it is reasonable to con-
clude that the target of prevention has
wrongfully imposed a dire risk on others.
The analogy with the law of conspiracy
and of self-defense takes us only so far, how-
ever, since it shows only that if similar con-
ditions were satisfied in the case of a state or
a terrorist group, it would be justifiable to
use force to arrest and punish them.
Whether it can be justifiable to use deadly
force against them depends on the addi-
tional assumption that the wrongful act they
plan to commit will be sudden and will cause
massive violations of human rights. In such
cases, the use of military force will often be
necessary to alleviate these threats.
One might object that the idea of alleviat-
ing an unjustly imposed risk through preven-
tive action justifies inflicting harm only on the
one who has imposed the risk, whereas pre-
the preventive use of force
ventive action will inevitably involve a risk of
harm to others. As we have already acknowl-
edged, preventive military action (unlike
ordinary acts of law enforcement) will almost
invariably risk harm to the innocent. This is
equally true, however, of intervention to stop
a presently occurring genocide and of inter-
diction, and even of the least controversial
cases of self-defense against an aggressive
attack. Unless one is willing to embrace an
absolute pacifist position that rules out the
use of military force in self-defense, the fact
that a preventive response to the wrongful
imposition of dire risk may involve harm to
innocent persons does not necessarily render
it morally illegitimate.
Given that military action typically
involves much greater risk of harm to inno-
cent persons than ordinary policy action
against conspirators, the threshold of
expected harm needed to justify preventive
military action should be high. From a cos-
mopolitan perspective, an appropriate
threshold is the occurrence of massive viola-
tions of basic human rights. As noted earlier,
by “basic human rights” we mean the most
widely acknowledged rights that are already
recognized in the major human rights con-
ventions. The requirement of massive viola-
tions of basic human rights sets a very high
threshold for preventive action.
Advocates of a blanket prohibition on
preventive force have one last arrow in their
quiver. They can argue that prevention car-
ries special risks that are not present in the
case of armed responses to actually occur-
ring attacks. The appeal to hypothetical
examples such as that of the sudden release
of a deadly virus, it might therefore be
argued, glosses over an important distinc-
tion between the justification of a particular
action and the justification of a rule allow-
ing that type of action. Even if preventive
action would be morally justified when cer-
7
tain highly ideal conditions are satisfied, it
does not follow that we should replace cur-
rent restrictions on preventive force with a
more permissive rule.
The prima facie moral argument for pre-
ventive action asserts that preventive action
may be undertaken to remove or mitigate a
wrongfully imposed dire risk. However, for a
dire risk to lead to severe harms, a long causal
series of events must typically be completed.
Even if the probability of each event in the
causal series is high, the probability of the
harm will be much lower, since it is the prod-
uct of all the probabilities of the events in the
series. Thus suppose that the causal series
anticipated by the potential preventer con-
sists of events A, B, C, and D and that each
event has a probability of .8. The anticipated
harm (the massive violation of basic human
rights) will occur only if the whole series of
events occurs; but the probability of this is
the product of the probabilities of each of the
members of the series: .4096. If the probabil-
ity of individual events is lower, the joint
probability is dramatically reduced. Further-
more, if events are uncertain––that is, prob-
abilities are unknown––it is impossible to
calculate precise probabilities.
It is not enough to show, the objector
would conclude, that there are some cir-
cumstances in which preventive action
would be morally justified. If these cases are
few—and if there is a significant risk that
those contemplating preventive action may
err in determining whether those circum-
stances obtain—then perhaps preventive
action should be prohibited altogether, as
traditional just war theory holds.11
This argument from the special risks of
prevention to a blanket prohibition is fatally
incomplete. It overlooks the fact that accept-
able risk reduction can be achieved not only
by a blanket prohibition but also by a more
permissive rule embedded in an appropriate
8 Allen Buchanan and Robert O. Keohane
institutional framework. Indeed, a blanket
prohibition comes at a high cost, since it
rules out action to prevent massive viola-
tions of basic human rights even when the
costs of prevention are very low and the like-
lihood of its success is very high.
A blanket prohibition rule also tends to
reduce the effectiveness of coercive diplomacy,
which may often be the most promising means
of preventing terrible harm as a result of fore-
seeable future aggression. Coercive diplomacy
can be defined as “bargaining accompanied by
threats designed to induce fear sufficient to
change behavior.”12 It includes deterrence but
is more comprehensive, since it also incorpo-
rates what Thomas Schelling once called
“compellance”––inducing an adversary not
merely to refrain from some action but to
engage in an action that it would not otherwise
undertake.13 For most states, most of the time,
illegality is viewed as a cost, if only because of
damage to reputation. If preventive force is
illegal, states will be less likely to use it and
efforts at coercive diplomacy through the
threat of preventive force will be less credible.
A Critique of the Legal Status Quo View
The Legal Status Quo (Security Council
authorization) view and the Cosmopolitan
Institutional view both reject the blanket pro-
hibition. Each holds that we can effectively
mitigate the special risks of prevention by
11
In one sense, this objection points out the obvious: that
tradeoffs exist in policy between timely action and cer-
tainty about the necessity of action. The longer one
waits, the more information one is likely to gain; but by
the same token, one’s ability to affect the situation may
be reduced. Britain and France had better reason in 1939
than in 1936 to regard Germany as a threat to their secu-
rity, but the cost of averting the threat was higher in 1939.
12
Paul Gordon Lauren, “Coercive Diplomacy and Ulti-
mata: Theory and Practice in History,” in Alexander L.
George and William E. Simons, eds., The Limits of Coercive
Diplomacy (Boulder, Colo.: Westview Press, 1994), p. 23.
13
Thomas C. Schelling, Arms and Influence (New Haven:
Yale University Press, 1966), pp. 69–78.
creating institutional safeguards. Although
reliance on Security Council authorization is
superior to a blanket prohibition, it is not the
best institutional alternative.
The first and most obvious moral flaw of
the legal status quo is the moral arbitrari-
ness of the permanent member veto. From a
cosmopolitan standpoint, there is no justifi-
cation for this radically unequal distribution
of decision-making authority. The veto seri-
ously impugns the legitimacy of the legal
status quo.
Second, the use and anticipated use of the
veto can block preventive actions that we
have morally compelling reasons to under-
take. As the Independent International Com-
mission on Kosovo argues, and as the
secretary-general of the UN acknowledged at
the time, the case for military action to pre-
vent (further) ethnic cleansing of Albanians
in Kosovo was very strong, but no resolution
to intervene was proposed in the Security
Council because it was known that Russia
(and perhaps China as well) would veto it.14
Third, the legal status quo lacks provi-
sions for holding the decision-makers who
propose preventive force (and those mem-
bers of the Council who must approve it)
accountable for their actions. There is sub-
sequently very little assurance that only
morally defensible interventions will be
authorized. Without standards for when
preventive action is justified and without
accountability mechanisms to increase the
probability that those standards are met,
authorization is likely to be unprincipled
rather than guided by a proper concern for
human rights.
These flaws constitute a strong prima
facie case for rejecting the Legal Status Quo
view—but only if a superior alternative is
feasible. We have seen that none of the three
conventional alternatives––the Just War
Blanket Prohibition, National Interest, and
the preventive use of force
Expanded Right of Self-Defense views––is
adequate. If the current legal status quo
ought to be abandoned, it must be for a new
arrangement that provides more, not less,
accountability.
The Risks of Preventive Action
Even though preventive military action is
sometimes justified, it creates a number of
serious risks, two of which are of special
concern with respect to preventive action,
even if they are not unique to it. The first is
that self-interest masquerading as concern
for the common good may lead to decisions
that are unjustifiable. The second is that pre-
ventive action will undermine existing ben-
eficial institutional norms constraining the
use of force. Each of these risks is particu-
larly worrisome since predictions that vio-
lence may occur are generally more subject
to error and bias than observations that it is
already occurring.
These risks indicate the need for institu-
tional safeguards to make decision-making
regarding the preventive use of force more
responsible. The more effective these safe-
guards are, the more the probability of abuse
will be reduced, and the more justified it will
be to undertake preventive action to cope
with massive violations of basic human
rights. By reducing the risks of abuse, insti-
tutional safeguards can make preventive
action more feasible on a greater range of
issues, promoting basic human rights.
Institutional safeguards therefore take the
sting out of the fact that there is no unique,
nonarbitrary threshold of probability of
harm needed to justify prevention. We can-
not reduce the risk of unjustified preventive
action to zero, but we can lower it suffi-
14
Independent International Commission on Kosovo,
Kosovo Report: Conflict, International Response, Lessons
Learned (New York: Oxford University Press, 2001).
9
ciently so that preventive action can become
a more useful tool of policy. Our institu-
tional analysis in the next section is designed
to show that this can be achieved.
COSMOPOLITAN INSTITUTIONS FOR
ACCOUNTABILITY
We advocate three principles for designing
institutions to govern the use of preventive
force. The first is effectiveness: the institution
should effectively promote the responsible
use of force to prevent massive violations of
basic human rights, regardless of whose
rights are threatened. One way to achieve
this is to design institutions governing the
preventive use of force in which decisions to
use force preventively are made by agents
who are, comparatively speaking, morally
reliable, and who are provided incentives to
make these decisions responsibly.
Only those states that have decent records
regarding the protection of basic human
rights should be allowed to participate in
institutional processes for controlling the
preventive use of force. Those who have
demonstrated respect for human rights over
a considerable period of time are generally
more reliable decision-makers regarding the
use of force for protecting human rights
than those who have not. We call this the
comparative moral reliability criterion.
Yet even agents that are morally reliable,
comparatively speaking, may act unreliably
unless they are provided with incentives that
induce them to make decisions on the basis
of the best available information, and to take
actions that are consistent with cosmopoli-
tan principles. In particular, incentives are
needed to counter the tendency of those
proposing the preventive use of force to
overestimate the risk to be averted and the
tendency of others to shirk responsibilities
to protect basic human rights.
10
Our second design principle also flows
from a core commitment of the cosmopoli-
tan ethical perspective: mutual respect for all
persons. Being willing to help protect the
basic human rights of all persons is one
important way in which we show respect.
Showing respect also entails being willing to
justify one’s actions to others. Arrogant dis-
missal of others’ views without providing
reasons and evidence is inconsistent with
this principle.
Within democratic societies, this princi-
ple of mutual respect means that people
wielding power and the means of coercion
must respond to queries about their use
from those who ultimately must authorize
these actions––members of the society in
question. Those who wield power must also
offer justifications of their actions to those
who are affected by them, including those
who are not members of their societies. In
the words of the Declaration of Indepen-
dence, they must show “a decent respect to
the opinions of Mankind.”
Our third principle of institutional
design, inclusiveness, reflects the values of
both effectiveness and mutual respect.
Effectiveness in world politics depends on
the independent actions of many states.
Therefore, the principle of effectiveness sug-
gests at least a presumption in favor of max-
imal inclusiveness, unless countervailing
considerations intervene. In addition, the
principle of mutual respect implies that
institutions should be as inclusive as is feasi-
ble, given their goals, since all are entitled to
equal regard.
In many cases effectiveness, mutual
respect, and inclusiveness will point in the
same direction. But it is worth noting that
there can be a conflict between promoting
effectiveness by selecting agents that possess
comparative moral reliability, and inclusive-
ness. Inclusiveness argues for giving all states
Allen Buchanan and Robert O. Keohane
access to participation in institutional
processes for controlling the preventive use
of force, whereas the need to enlist agents
who have a sincere commitment to and a
sound understanding of human rights
favors restricting participation. How the
tradeoffs between inclusiveness and com-
parative moral reliability are made is of cru-
cial importance for institutional design, as
indicated by our discussion below of possi-
ble arrangements for a democratic coalition
to supplement the authorization processes
of the Security Council.
The Concept of Accountability and the
Preventive Use of Force
No single state can be counted upon fully to
take into proper account the interests of
others, particularly when considering the
use of military force. Therefore, processes by
which states, or nonstate actors, can hold
states accountable are essential if the
parochial concerns of the most powerful
states are not to be allowed to prevail over
broader interests and shared values.
Standard definitions of accountability
emphasize both information and sanc-
tions. “A is accountable to B when A is
obliged to inform B about A’s (past or
future) actions and decisions, to justify
them, and to suffer punishment in the case
of eventual misconduct.”15 Accountability
operates both ex ante and ex post. Those
states that propose to use force preventively
must, under cosmopolitan principles, con-
sult with other states and make their inten-
tions known to international society more
generally before using force. Having used
force, they must provide information,
answer questions, and subject themselves
to sanctions according to rules that have
been established in advance.
Inclusion of sanctions (including punish-
ments) is crucial, since there will be little
the preventive use of force
incentive to comply with the requirement to
answer questions and provide justifications
unless those to whom the action is to be jus-
tified can sanction those in power. In addi-
tion, accountable institutions require
specification of the standards to which
agents are to be held accountable, those to
whom they are to be accountable, and a will-
ingness on the part of agents to provide
information and answer questions about
their behavior.
For a cosmopolitan view, the most funda-
mental standard of accountability is that
states must act in ways that are designed to
respect and protect the human rights of all
persons. They are only to use force or to
authorize its use for the sake of preventing
violations of human rights. In addition, jus-
tified preventive action must include stan-
dards for how preventive force is to be
applied so that it reflects the cosmopolitan
commitment to basic human rights.
Prominent among the substantive stan-
dards for determining the way in which pre-
ventive action may be conducted will be the
traditional jus in bello principles––that
harm to innocents should be minimized,
force must be proportional to the end to be
achieved, excessive force is to be avoided,
and unnecessary suffering should not be
inflicted on enemy combatants. To institu-
tionalize adequately these standards
requires mechanisms that help to ensure
that they are applied correctly and that there
are effective sanctions to increase compli-
ance with them.
15
Andreas Schedler, “Conceptualizing Accountability,”
in Andreas Schedler, Larry Diamond, and Marc F. Plat-
tner, eds., The Self-Restraining State: Power and Account-
ability in New Democracies (Boulder, Colo.: Lynne
Rienner, 1999), p. 17. The word “punishment” could be
interpreted in too formal and restrictive a way. We
accept this definition only with the proviso that it refers
more broadly to a penalty.
11
 To determine whether these standards are
met, information is required. Thus the sec-
ond component of a defensible accountabil-
ity system is provisions for the sharing of
information. A state proposing to use force
must share its information about the risks of
large-scale human rights violations that
would result from inaction as well as the
risks of action, especially regarding harm to
innocent persons. The information that it
provides must be comprehensive enough for
other states to make independent judgments
on the costs and benefits of the preventive
use of force.
Procedures that ensure information shar-
ing are necessary for providing the appro-
priate incentives for action. Without such
procedures, states could pretend to be using
preventive force to protect human rights
while in fact acting to preserve their hege-
mony in a region, to protect weak regimes
with which they are allied, or to gain eco-
nomic advantages. The requirement of
information sharing is designed, therefore,
not only to correct false beliefs but also to
reduce opportunistic use of force. Informa-
tion sharing is therefore required by the
design principle of effectiveness. It is also
required by the principle of mutual respect,
since mutual respect is evidenced by a will-
ingness to share information with equals
rather than to demand deference from those
regarded as inferior in the capacity to under-
stand issues and make decisions.
Introducing sanctions for the violation of
standards is the third element of an account-
ability system. It is also the most difficult to
institutionalize. Different societies hold dif-
ferent values, and states are likely to privilege
their own interests, and it is therefore infea-
sible to rely on voluntary compliance. World
politics is often seen as a “self-help” system
in which each state is responsible for its own
security.16 Without sanctions, rules for the
12
regulation of the preventive use of force
would almost certainly be ineffective. Unless
there are sanctions for violations of the
requirement to share information, some
states will misrepresent the facts, exaggerat-
ing the probability of the harm that they
propose to prevent. Unless there are sanc-
tions against those who use excessive force,
states are likely to discount the harm their
forceful actions will inflict on others. With-
out effective sanctions, the institutions will
be ineffective, since in the absence of sanc-
tions, “the strong do what they can and the
weak suffer what they must.”17
Standards, information, and sanctions are
all crucial to accountability, but they leave
open the question, To whom should agents
who are considering wielding military force
be held accountable? We have already sug-
gested that a cosmopolitan view entails a
presumption in favor of inclusiveness. In
addition, the design principle of mutual
respect creates a presumption that partici-
pation in institutional processes for the use
of force to protect the human rights of all
should be open to all. This presumption
could be rebutted, however, if the effective-
ness of the accountability regime requires
restricting participation to those agents that
are comparatively morally reliable.
Ex Ante Accountability
With respect to a state’s proposal to engage
in the preventive use of force, ex ante
accountability requires that all of the issues
and options (including nonmilitary
options) be discussed, and that states that
question the necessity of the intervention
16
Kenneth N. Waltz, Theory of International Politics
(Reading, Mass.: Addison-Wesley, 1979).
17
Thucydides, The Peloponnesian War, Richard Crawley,
trans., bk. V, para. 84 (New York: Modern Library, 1982),
p. 351.
Allen Buchanan and Robert O. Keohane
have the opportunity to interrogate those
who propose it. Consistent with the princi-
ple of mutual respect, rational persuasion
must be employed. Reliance on principled
rational persuasion also means that parties
speaking in favor of or against preventive
military action must acknowledge that their
decisions will have precedential value for
future decisions. Both the party contemplat-
ing intervention and those who must ratify
its decision must therefore acknowledge at
least a prima facie obligation, on future
occasions, to follow the principles that they
have invoked to defend their actions, regard-
less of the identities of the intervener and
the target of intervention. In this sense, the
expectation that current decisions will have
precedential value for future decisions
serves as a kind of veil of ignorance, making
the procedure conducive to fairness and
therefore to mutual respect.18 Even if a very
powerful state may not expect to be a target
of future preventive action, it may worry
that the principles it employs to justify or to
oppose preventive action will be used by
other powerful states to justify their own
preventive uses of force.19
Mutual respect also requires that all par-
ticipants in the process of deciding on the
preventive use of force must have equal
standing to pose questions to the potential
intervener and offer arguments against
intervention. In addition, there is at least a
strong presumption against according
weighted votes or veto power.
Finally, the process of deliberation ex
ante must involve the participation of a
group of states with diverse interests. Too
much uniformity of interests among the
participants would run the risk of bias, per-
haps even turning the regime for accounta-
bility into a tool for domination. Protection
of the human rights of all requires diversity
of perspectives.
the preventive use of force
Ex Post Accountability
It is quite possible that what is discovered
after even a successful military attack will
not justify the actions that were taken. Ex
ante accountability is therefore not suffi-
cient; provisions for ex post accountability
are also essential. The attacking states must
come back with a full report to the body that
authorized their actions. They must also
allow an impartial commission appointed
by that body to have free and timely access
to all places in the country controlled by the
attacking states in the course of the preven-
tive action. That is, they must facilitate the
generation and publicizing of the best avail-
able impartial information about the actual
effects of preventive use of military force.
Evaluation of the results of the preventive
military action would focus on the consis-
tency of the acting states’ behavior with the
statements they made in the ex ante
accountability process. Two issues are par-
ticularly important:
Was the information gained ex post about
the risk-imposing actions of the target con-
sistent with the statements made by the
states proposing action ex ante?
Were the military actions of the attacking
states consistent with their assurances ex
ante that their actions would be propor-
tional to the objectives being attained? The
justification of military action on the basis
18
For a more general argument, see Robert O. Keohane,
“Governance in a Partially Globalized World,” American
Political Science Review 95, no. 1 (March 2001), pp. 1–13.
See also Geoffrey Brennan and James M. Buchanan, The
Reason of Rules: Constitutional Political Economy (New
York: Cambridge University Press, 1985), p. 30; and Nor-
man Frohlich and Joe A. Oppenheimer, Choosing Justice:
An Experimental Approach to Ethical Theory (Berkeley:
University of California Press, 1992).
19
In the World Trade Organization states behave in this
way. They are reluctant to put forward legal arguments
to defend their own practices that could be used in the
future by other states to justify protectionist measures.
13
of risk only justifies action necessary to
remove or significantly reduce the risk. It
should ultimately be up to the authorizing
body, not to the attacking states, to decide
how far military action needs to go to attain
the agreed objectives.
With respect to the justification of an
attack, two contrasting inferences might be
drawn from information gathered after the
war: one, on the whole, the ex post investi-
gations support the assessments of risk
made before the attack; or two, on the
whole, the ex post investigations fail to cor-
roborate those assessments. Of course,
there would be gradations of judgment
between these two poles.
If the situation corresponds more closely
to situation one, the attacking states would
indeed have performed a public service for
the world by eliminating or reducing the
threat that weapons of mass destruction
would be used or that large-scale violations
of basic human rights would be inflicted by
some other means. These positive contribu-
tions would be counted against their finan-
cial obligations. Indeed, it would be
presumed that those states that had not
shouldered the risk of preventive military
action would bear special responsibility for
financial support in rebuilding the target
country should the preventive use of force
have caused extensive damage. Insofar as
feasible, they would also bear responsibility
for peace enforcement. That is, under these
conditions those states that had not sup-
ported preventive military action would be
sanctioned as “free riders,” who were
informed about the threat but refused to act
in a timely manner. Although the economic
obligations of such free riders would be lim-
ited by their economic capabilities––poor
and weak states would not be unduly bur-
dened in any case—they would, in this situ-
ation, be greater than those imposed on
14
states with comparable resources that sup-
ported the military action. 20
However, if the ex post review concludes
that the preventive action was unjustified,
the attacking states and their supporters
would be held accountable. They would
have to face sanctions for their actions in
much the same way that states participating
in the international regime are liable for
their infringements of its rules. They would
be required to compensate those who suf-
fered harm from the preventive action and
to provide full financial support for opera-
tions that restore the country’s infrastruc-
ture and enable it to govern itself effectively.
Furthermore, the intervening parties would
not be allowed to control the political situa-
tion in the conquered country, or to deter-
mine the allocation of aid or the awarding of
contracts to firms offering services for the
reconstruction effort. If states know ex ante
that these rules are in place, incentives for
opportunistic interventions aimed at domi-
nation rather than protection of human
rights will be diminished.21
A crucial component of our proposal is
an impartial independent commission,
appointed by the authorizing body (such as
the Security Council). This commission
would have prompt access to the conquered
20
The analogy of conscientious objection may be help-
ful. Conscientious objectors may avoid military service
but are expected to subject themselves to at least equiv-
alent burdens and risks as those who do carry arms, by
serving as medics, for example.
21
The Fourth Geneva Convention relative to the Protec-
tion of Civilian Persons in Time of War of August 12,
1949, and other provisions of international law specify
the obligations of occupying powers. These include
ensuring respect for basic human rights, including
restoration and maintenance of law and order and pro-
vision of food and medical care to the population. Under
our proposal these obligations would not be diminished,
but occupying powers whose actions had been judged by
the ex post review to be unjustified would have their abil-
ity to make decisions severely constrained.
Allen Buchanan and Robert O. Keohane
country and would determine the magni-
tude of reconstruction aid needed. It would
also determine the allocation of responsibil-
ities to provide the necessary funds.
One particularly attractive feature of the
ex post accountability requirement is that it
provides the intervening state with a power-
ful incentive to comply with the just war
principle of using the minimum of force suf-
ficient to achieve the goal of prevention. This
requirement places the burden of evidence
on the intervening state: it must be able to
show that its ex ante judgments in support of
preventive action were valid, and this
requires that evidence to substantiate those
judgments be available for all to see ex post.
Were the intervener to use excessive force,
this could destroy the needed evidence. Ex
post accountability thus institutionalizes an
important just war principle in such a way as
to create incentives for compliance with it.
Similarly, the knowledge that its actions
will be subject to an impartial review would
give the intervening state an incentive to
minimize harm to noncombatants, espe-
cially if significant penalties would follow
upon a finding that the harm to noncom-
batants was excessive. These would prima-
rily consist of compensation to those
harmed or their families, but might include
other sanctions as well.
The key to this mechanism of accounta-
bility is a contingent contract that applies
both to states seeking to use force preven-
tively and to states opposing the preventive
use of force. States would agree that after the
conflict they would abide by the judgments
of the authorizing body on the validity of
their ex ante arguments. They would also
agree that, should this judgment go against
them and the independent commission
determine a level of compensation, they
would pay such compensation (subject per-
haps to an appeals procedure). It should be
the preventive use of force
emphasized that states opposing preventive
action would also be liable to pay compen-
sation if ex post discoveries confirmed the
claims made ex ante by the states advocating
preventive action and disconfirmed their
own ex ante objections. In other words, the
contract is double-sided and does not dis-
criminate against potential interveners.22
Such a contractual arrangement would
decrease the incentives for dishonesty on the
part of both proponents and opponents of
preventive action in their ex ante presenta-
tion of evidence. A major problem facing all
but the most powerful states, and all publics,
is that they do not have access to privileged
intelligence information, particularly with
respect to closed societies and dictatorships
such as North Korea and Iraq. They are
therefore subject to being deceived about
the real state of affairs. And since they are
aware that they do not have access to much
information, they may become unreason-
ably skeptical of proposed interventions.23
Another advantage of this contractual
arrangement is that it would limit abuses of
victory in a preventive war. When interven-
tions result in large-scale damage, decisions
about how to reconstruct the country,
including those relating to its political-eco-
nomic system and control of its resources,
must be made in consultation with the
22
Here it might be objected that the strongest states
might agree ex ante to pay compensation if an ex post
review found their preventive actions to be unjustified,
but simply fail to pay up. To avoid this problem, assets
could be escrowed ex ante. Such an arrangement was
used in the agreements that ended the U.S.-Iran hostage
crisis in 1981 and that established the U.S.-Iran Claims
Tribunal. See Robert Carswell and Richard Davis,
“Crafting the Financial Settlement,” in Warren Christo-
pher et al., eds., American Hostages in Iran: The Conduct
of a Crisis (New Haven: Yale University Press, for the
Council on Foreign Relations, 1985), pp. 201–34; docu-
ment in ibid., Appendix D, pp. 405–22.
23
See George Akerlof,“The Market for Lemons,” Quarterly
Journal of Economics 84, no. 3 (August 1970), pp. 488–500.
15
authorizing body. When the states engaged
in preventive military action have succeeded
in their objective—the elimination of the
risks for which they claimed to go to war—
they must subject their plans for postwar
reconstruction and prevention of recur-
rence of the risks to the authorizing body.
All states could stand to gain from the imple-
mentation of the accountability safeguards that
we have described. States not expecting to use
preventive military force would gain more reli-
able information before having to make deci-
sions on its authorization by a collective body
such as the United Nations. States expecting to
use preventive force could hope to gain credi-
bility for their claims, and therefore be likely to
generate broader support for preventive mil-
itary actions that are justified. And all could
benefit from the combination of induce-
ments to avoid dysfunctional deadlock and
abuses of authority.
ALTERNATIVE INSTITUTIONAL
MODELS FOR ACCOUNTABILITY
In this section we discuss three alternative
arrangements that might reasonably satisfy
the requirements of cosmopolitan accounta-
bility. None of these arrangements may be
politically feasible at present, even though we
have sought to design them in a way that takes
a nonideal world of primarily self-interested
states as a given. We hope that people of good-
will and sophistication may be interested in
refining and implementing them, or institu-
tions inspired by them. These suggestions
therefore fall somewhere between mere
thought experiments and proposals that could
be implemented under current conditions.
Institutional Model 1:
Accountability without the Veto
The first model relies exclusively on the
Security Council but creates mechanisms
16
for ex ante and ex post accountability and
removes decisions about preventive force
from the scope of the Council’s permanent
member veto. The veto could still be exer-
cised in other Council deliberations, but it
would be abolished in the case of decisions
whether to use preventive force.
On this model, the provisions for ex ante
accountability described above would struc-
ture the Council’s deliberations, if time per-
mitted. If nine members of the Council
voted in favor, a proposed preventive action
would go forward, subject to an impartial ex
post review as described. The requirement of
a supermajority of nine votes appears rea-
sonable, since it ensures that a diverse group
of states would have to agree to the action.
If a state or group of states uses preventive
force prior to Council authorization on the
grounds that there was no time to seek it, then
they would also be subject to an impartial ex
post review.Such a review would also determine
whether it had been reasonable to go forward
without prior approval by the Council. Should
the review determine that they were not justi-
fied in attacking without prior approval, they
would be subject to sanctions at least as severe
as those imposed on attackers who received
authorization under false pretenses.
There are three major arguments in favor
of dropping the veto for these decisions. First,
it has no ethical standing, since it is a political
artifact of the era in which the UN was
founded, and offends against the presump-
tion of equality implied by the design
principle of mutual respect by giving dis-
proportionate power to some states. Second,
removing the veto reduces the inertia in the
present system, making it more likely that
force could be used preventively when its use
would be morally compelling. Third, abolish-
ing the veto would encourage states that resist
the use of preventive force to propose con-
structive alternatives. The lack of a veto would
Allen Buchanan and Robert O. Keohane
make it necessary for states to persuade others
that the preventive use of force is not neces-
sary. In some cases this would require that
they present alternatives to its use. At present,
those who wield the veto need not persuade,
they can simply dictate the outcome.
Despite these attractions, any proposal that
includes abolishing the permanent member
veto will meet insuperable political resistance
to its implementation in the foreseeable
future. States that now possess the veto are
unlikely to relinquish it, and no one else can
make the proposed change. Moreover, it is
especially unlikely that they would relinquish
the veto over something as important and
controversial as the preventive use of force.
Even if the permanent members saw some
merit in the proposal to relinquish the veto
only in decisions concerning preventive use of
force, they might fear a slippery slope leading
to the loss of the veto in other areas. Finally,
there is a danger that removing the veto for
states wielding significant force could create
military conflict between the United Nations
and a great power, which would be a disaster.24
This alternative for institutional reform, then,
is the least feasible of the three alternatives.
Institutional Model 2:
Accountability despite the Veto
A less radical and correspondingly more fea-
sible model would create mechanisms for ex
ante and ex post accountability as in model 1
but leave the permanent member veto in
place.25 On this second model, the Security
Council would have to approve military
action by the procedures currently in the
Charter. The Council would appoint an
impartial body to determine whether the
intervener’s ex ante justification for preven-
tive action is confirmed ex post.
Such an arrangement might reduce the like-
lihood that vetoes would be used, since those
states reluctant to endorse preventive action
the preventive use of force
would at least be assured that the results would
be impartially reported, with provisions for
penalties in the case of malfeasance. In other
words, they would not be issuing a “blank
check”to those initiating force preventively. To
that extent, the second model’s provisions for
accountability would alleviate the problems
that the veto poses. However, it should be
emphasized that for ex post accountability to
work, it would also be necessary for the Coun-
cil to create suitable penalties to be applied in
the case of a negative ex post evaluation.26
24
I. L. Claude likens the UN Security Council veto to “a
fuse in an electrical circuit,” reflecting the wise convic-
tion that “in cases of sharp conflict among the great
powers the Council ought, for safety’s sake, to be inca-
pacitated.” I. L. Claude, Power and International Rela-
tions (New York: Random House, 1962), p. 160.
25
Article 51 of the UN Charter states that Members of
the United Nations that respond with force to an armed
attack shall “immediately” report their actions to the
Security Council. Our institutional model 2 goes sig-
nificantly beyond Article 51 in three ways: it expands the
scope of a reporting requirement beyond the case of
self-defense against an actual armed attack to cover pre-
ventive force; it specifies that the report must be made
by an impartial body and must explicitly address issues
of fact and of reasonableness in the justification of the
use of force; and it attaches penalties in the case of an
unfavorable report. Article 51, in contrast, allows the
party using force to report on itself, specifies no criteria
of evaluation for the report, and mentions no conse-
quences of an unfavorable report.
26
This necessity raises the problem of an ex post veto by
the intervening state or states of the findings of the
impartial commission or penalties implied by those
findings. This problem could be solved by adapting a
suggestion recently made by Thomas Franck, who
points out that Article 27 provides that “decisions of the
Security Council on procedural matters shall be made
by an affirmative vote of nine members.” That is, on
procedural matters, the veto is inapplicable. Ex ante the
Council could decide that votes on the composition
and report of the impartial commission, and on any
recommendations for penalties, would be considered
procedural. (Professor Franck’s proposal concerns
specification of what constitutes a “material breach”
regarding UN resolutions concerning Iraq. See his
“Inspections and Their Enforcement: A Modest Pro-
posal,” American Journal of International Law 96 (Octo-
ber 2002), pp. 899–900.
17
Model 2, like model 1, builds on the existing
UN structure, but is more feasible than
model 1, since it does not require those with
the veto to relinquish it.
Nevertheless, even the measures pro-
posed in model 2 might seem to be too
costly for the leader of a coalition for pre-
vention. Why should the coalition leader
accept these constraints if it is to bear the
principal costs of military action? The
answer is that only by accepting con-
straints, ex ante, could the coalition leader
make credible its justification for preven-
tive force and its own promises regarding
its conduct during and after the conflict.27
And credible promises are essential to
induce other members of the Security
Council to grant authorization for the use
of preventive force. Furthermore, securing
a broad coalition ex ante can provide valu-
able insurance for the interveners, since
they could otherwise be stuck with bearing
the full cost of reconstruction if political
conditions in the occupied territories
worsen after a successful military opera-
tion. States that do not consent to the use of
force in the first place are likely to demand
a high price for being involved in the post-
conflict reconstruction.28
Although superior to model 1, model 2 is
also unsatisfactory. It avoids the impracti-
cality of model 1, but keeping the permanent
member veto comes at a high price, because
the veto is both morally arbitrary and facili-
tates a lack of accountability. The veto gives
those who wield it little incentive to per-
suade others by principled argument or to
offer constructive alternatives to the use of
force. So long as the decision whether to use
preventive force rests exclusively with the
Security Council as currently constituted,
there will be no recourse against deadlock
produced by arbitrary and self-interested
use of the veto.
18
Institutional Model 3:
A Role for a Democratic Coalition
The third and most defensible institutional
alternative is to implement the Security
Council–based accountability mechanisms
described in model 2 but supplement them
with a supporting role for a coalition of rea-
sonably democratic states. By democratic
states we mean those with constitutional,
representative governments, competition
for elected positions through reasonably fair
elections, and entrenched basic civil and
political rights. A fairly wide range of polit-
ical institutions and cultures can meet these
minimal requirements. There is no sugges-
tion that democratic states include only U.S.
or Western European–style democracies.
The point is simply to exclude states that are
unambiguous violators of human rights.
Of course there is disagreement as to pre-
cisely how these elements of democracy are
to be understood. For our purposes, how-
ever, there is no need to enter into that
debate. Our proposal is to begin with a core
group of states whose democratic creden-
tials are uncontroversial, such as the mem-
bers of the European Union plus states such
27
When the United States invaded Iraq in 2003 without
Security Council authorization, the Bush administra-
tion’s chief justification for doing so was to destroy
weapons of mass destruc1tion, which it said were pres-
ent in that country. It can be argued that as of seven
months after the invasion began, the United States has
failed to show either that there were weapons of mass
destruction in Iraq at the time of the invasion or that
there was good reason to believe that they were present.
If this still appears to be the case after all the evidence is
in, the problem of achieving credibility for the preven-
tive use of force will be even greater in the future. The
case for the ex ante and ex post accountability mecha-
nisms we propose will be all the stronger.
28
The United States gained Security Council approval
for the first Gulf War in 1991, but not for the invasion of
Iraq in 2003. Other countries paid approximately 80
percent of the cost of the former conflict; so far, the
United States is paying over 80 percent of the costs of
the latter.
Allen Buchanan and Robert O. Keohane
as Australia, Canada, Chile, Costa Rica,
Japan, South Africa, and South Korea.
Unlike NATO, the coalition that intervened
in Kosovo in 1999, such a coalition would be
open to states from all regions of the world.
Indeed, we expect that most of its members
would not be from Europe or North Amer-
ica. The initial members would be author-
ized to admit additional countries through a
transparent process utilizing publicly stated
criteria for membership. Consistent with
our overall argument, the democratic coali-
tion would itself need institutionalized pro-
cedures for accountability, both with respect
to admission to membership and for sub-
stantive decisions.
From a principled standpoint, creating a
democratic coalition is attractive because it
would reinforce the role of states that meet
our standard for comparative moral reliabil-
ity. This is not to say that we believe that
democracies always, or even typically, act in
ways that are consistent with cosmopolitan
morality. Indeed, a key premise of our insti-
tutional proposals is that all states often
behave in parochial and self-interested ways.
We believe, however, that when democracies
violate cosmopolitan principles, they are
more likely to be criticized by their citizens
for doing so, and will be more likely to rectify
their behavior in response. They may not be
morally reliable agents, but comparatively
they are more reliable than autocracies.
The third model is feasible, since formal
UN action would not be necessary to imple-
ment it. Indeed, if a large and diverse cluster
of democracies proposed it, no single state
could easily block it. The democratic coali-
tion would be based on agreements among
its members—not necessarily through a for-
mal treaty. Over time, its practices could
become part of customary international law.
Furthermore, recent experience with respect
to the Anglo-American war against Iraq sug-
the preventive use of force
gests that democratic countries are capable
of making independent judgments about
the proposals of a superpower, even when
faced with the prospect of bilateral sanc-
tions. Chile and Mexico, for instance,
refused to support the resolution authoriz-
ing war against Iraq put forward by the
United States and United Kingdom in Feb-
ruary and March of 2003.
In international law, recognition of new
states is awarded by existing states, acting
individually or through institutions such as
the United Nations. Those entities that are
indisputably sovereign states have the
authority to determine which other entities
will be treated as sovereign. Likewise, the
initial core of the democratic coalition
would be self-designated states whose cre-
dentials as stable democracies are unassail-
able. They would then develop membership
criteria that would determine which other
states would be eligible. These criteria would
be public, developed and applied through
transparent, accountable processes. An
additional advantage of a democratic coali-
tion is that accountability is much more
institutionalized within democratic soci-
eties than in authoritarian ones.
The incentives for joining the democratic
coalition would be substantial.
Submitting to the coalition’s delibera-
tions could enable states proposing the use
of military force to gain legitimacy for their
actions without Security Council authori-
zation. Furthermore, the accountability
arrangements of the coalition would
increase the credibility of its arguments,
thereby enabling it to gain allies who would
share the costs of military action and recon-
struction. And other members would
receive valuable recognition as democratic
states that are regarded as sufficiently trust-
worthy to participate in important deci-
sions regarding the use of force. More
19
important, they would gain decision-mak-
ing authority, both with respect to legit-
imizing the preventive use of force and for
determining which other states should be
allowed to join. There would also be a pow-
erful incentive to help initiate the coalition
or, if others initiated it, to join early, since
the original members would have more say
in determining the rules under which the
coalition would operate.
The principle of mutual respect implies
that the rich members of the democratic
coalitions would have to offer poorer
democracies better treatment than is now
the case with respect to both trade and aid.
Members of the democratic coalition would
extend trade privileges to one another,
beyond the requirements of the World Trade
Organization, and the richer members
would offer particularly generous aid to the
poorer ones.29
The democratic coalition would not
replace the Security Council. Instead, if a
state or group of states proposed preventive
action to the Council and was unable to gain
its authorization, it could then present its
case to the democratic coalition. The demo-
cratic coalition would have its own ex ante
and ex post accountability processes that
might differ in detail from those of the Secu-
rity Council under model 2, but that would
likewise be designed to approximate as
closely as feasible the design principles for a
cosmopolitan accountability regime.
The third model, then, includes two parts.
The first is the existing Security Council
arrangement, including the permanent
member veto, but importantly modified to
include ex ante and ex post accountability
mechanisms. The second is a distinct body
for decisions concerning the use of preven-
tive force, consisting of a coalition of demo-
cratic states, geographically unrestricted in
membership, and designed to go into oper-
20
ation only in the event of deadlock in the
Security Council.
The most obvious advantage of model 3 is
that it provides for the possibility of respon-
sible decisions to use force when the Secu-
rity Council fails to do so. But another,
equally important advantage is that the pos-
sibility that a decision will go to the demo-
cratic coalition provides an incentive for the
Security Council to act more responsibly.
Model 3, then, remedies the chief deficiency
of model 2 without requiring the impracti-
cality of abolishing the permanent veto as in
model 1. If model 3 were adopted, perma-
nent members of the Council would be
more reluctant to veto proposals for preven-
tive action without good reasons and sub-
stantial support, if doing so would transfer
decision-making authority to an arena in
which they did not enjoy veto power—and
in which they might perhaps not be repre-
sented at all.
Conversely, the continued operation of
the Security Council, with strengthened
institutions for accountability of the sort we
recommend, could make the idea of a dem-
ocratic coalition more attractive. It would
reduce the likelihood that resort to the dem-
ocratic coalition would be overused, creat-
ing a schism between the democracies and
the other members of the UN.
The third institutional model creates
healthy competition with the UN system
without bypassing it altogether. More
specifically, the democratic coalition pro-
vides an incentive for the permanent mem-
bers to use the veto more responsibly and for
all members of the Council to realize that
they no longer enjoy an absolute monopoly
on the legitimate authorization of preven-
tive use of force.
29
We are indebted to Michael Doyle for the suggestion
that led to this proposal.
Allen Buchanan and Robert O. Keohane
 None of these schemes could function
effectively without prior agreement on a
threshold criterion for predicted harm to be
prevented. We have argued that a starting
point for such a threshold is the idea of mas-
sive violations of basic human rights. To cre-
ate a workable criterion would require
gaining agreement among the participants
in these institutions on a minimal list of
basic human rights. The problem of specify-
ing when “massive” violations of basic
human rights have occurred would be more
difficult. There would, however, be clear
cases—for example, where the threatened
harm is the detonation of a nuclear bomb in
a densely populated area. A solution to this
problem would presumably involve a com-
bination of prior agreement on levels of risk
sufficient for triggering serious considera-
tion of preventive action and a developing
case law that would provide further speci-
ficity over time.
The concept of accountability is crucial in
thinking about more adequate institutional
procedures for decisions concerning the use
of force. States advocating preventive war
should be subjected both to ex ante and ex
post accountability. Ex ante, they must be
able to persuade other states of the merits of
their case for military action. Ex post, their
actions must be subjected to scrutiny and
potential sanctions.
Our view might be seen as unduly con-
straining to states that might otherwise
engage in justified preventive intervention.
This is indeed a possibility, but it must be
emphasized that our argument is compara-
tive. Our system for cosmopolitan account-
ability is less constraining than what we take
to be the most plausible of the competing
views, the Legal Status Quo view, since our
proposal merely envisages a democratic
coalition as an additional channel for
authorization.
the preventive use of force
Even the most powerful states wish others
to view their actions as legitimate. They
need allies when undertaking military
action and supporters to provide policing
assistance, civilian infrastructure, and finan-
cial support in the aftermath of war. By
meeting the requirements of the cosmopol-
itan accountability regime, a state proposing
preventive military action would gain legit-
imacy by submitting to the rule of law, and
credibility by allaying the suspicions of
other states that the justification given for
prevention was not sound or that preven-
tion was a pretext for conquest. The cosmo-
politan accountability regime should be
attractive, then, not only to those wishing to
constrain states bent on preventive action
but also to those seeking to engage in it.
This is not to say that we regard our pro-
posal as feasible in the short run. There is, as
Machiavelli noted, “nothing more difficult
to take in hand, more perilous to conduct, or
more uncertain in its success, than to take
the lead in the introduction of a new order
of things.”30 Considerable discussion and
refinement would be necessary for our
accountability regime to be accepted by
states. Countries in the global South will
have to be reassured that the rich countries
are really following cosmopolitan principles
of mutual respect. Potential interveners will
have to come to recognize that the advan-
tages of accountability outweigh constraints
on their freedom of action.
Even the process of discussing accounta-
bility in this way would be an advance, by
focusing attention on the problem of elicit-
ing accurate information in situations of
potential military conflict.31 Such discus-
sions could themselves enhance the reputa-
tional and credibility costs of ex ante
30
Niccolo Machiavelli, The Prince (1532), ch. VI.
31
We are indebted to Jeremy Waldron for this point.
21
dissimulation. Over time, we hope that our against its abuse. Establishing an institu-
proposal could provide an intellectual tionalized system of accountability for pre-
resource for reformers seeking to promote ventive war would constitute a progressive
effective multilateral action, with safeguards step in international governance.

22 Allen Buchanan and Robert O. Keohane